🗿
AI EngineeringJavaScript

Caveman

by DietrichGebert

Caveman is an AI Engineering skill for Claude Code, published by DietrichGebert in ponytail.

101.7K stars5.6K forkson DietrichGebert/ponytailAdded 2026/08/13Repository updated 2026/08/07
agent-skillsai-agentsclaudeclaude-codeclaude-code-plugincursor-rulesdeveloper-toolsllmprompt-engineeringyagni
Install in seconds
Install Caveman
Copy Caveman into your Claude Code skills folder. Run the command in your terminal, or review the source on GitHub before installing.
terminal
npx degit https://github.com/DietrichGebert/ponytail/tree/main/benchmarks/arms ~/.claude/skills/arms

Requires Node.js. Downloads this skill only — not the rest of the repository — into your Claude Code skills folder.

Without Node.js

git clone https://github.com/DietrichGebert/ponytail.git

Clones the whole repository, then copy the skill’s own directory into your skills folder yourself.

In this catalog

Source file
benchmarks/arms/caveman-SKILL.md in DietrichGebert/ponytail
Installs to
~/.claude/skills/arms
Collection
One of 13 skills cataloged from this repository
Category
AI Engineering2451 skills

What Caveman does

Caveman activates an ultra-compressed communication mode in Claude Code that cuts token usage by about 75% while preserving full technical accuracy. Use it when a user requests caveman mode, wants fewer tokens, or asks for brief responses.

Caveman is cataloged under AI Engineering on DirSkills. Caveman comes from a repository tagged agent-skills, ai-agents, claude, claude-code and claude-code-plugin.

Documentation

README

Respond terse like smart caveman. All technical substance stay. Only fluff die.

Persistence

ACTIVE EVERY RESPONSE. No revert after many turns. No filler drift. Still active if unsure. Off only: "stop caveman" / "normal mode".

Default: full. Switch: /caveman lite|full|ultra.

Rules

Drop: articles (a/an/the), filler (just/really/basically/actually/simply), pleasantries (sure/certainly/of course/happy to), hedging. Fragments OK. Short synonyms (big not extensive, fix not "implement a solution for"). Technical terms exact. Code blocks unchanged. Errors quoted exact.

Pattern: [thing] [action] [reason]. [next step].

This is the opening of the README. Read the full README on GitHub.

Commands Caveman provides

Slash commands named in this skill’s SKILL.md, listed in the order they first appear.

  • /caveman

Frequently asked about Caveman

  • What else does DietrichGebert publish alongside Caveman?

    Caveman is one of 13 skills that DirSkills catalogs from DietrichGebert/ponytail, the repository it ships in. Its siblings there include Ponytail, Ponytail and Ponytail Audit. Each one is a separate skill with its own page in this directory, installs the same way Caveman does, and is maintained by DietrichGebert in that same repository. The rest of the collection is listed on the DietrichGebert/ponytail page.

  • How does Caveman compare to other AI Engineering skills?

    Caveman ranks #27 by stars among the 2451 AI Engineering skills in this catalog. The most-starred ones next to it are Architecture Decision Records, AI-First Engineering and Agentic OS. DirSkills ranks by the star count of the repository each skill ships in, so that order reflects how popular those repositories are rather than any review of Caveman against them. Open each page to compare what they document and how they install.

More from DietrichGebert/ponytail

Caveman is one of 13 skills cataloged on DirSkills from DietrichGebert/ponytail.

See all 13 skills
🦥
3w ago

Ponytail

Ponytail puts an agent into lazy senior developer mode that enforces YAGNI, stdlib-first, and minimal-diff choices for any coding task. Use it to write, refactor, fix, or review code without over-engineering.
AI Engineering
101.7K5.6K
🦥
3w ago

Ponytail

Ponytail forces the laziest solution that actually works, favoring simplicity, the standard library, native platform features, and minimal diffs. Use it on any coding task to question scope, avoid over-engineering, and ship the shortest correct fix.
AI Engineering
101.7K5.6K
✂️
3w ago

Ponytail Audit

Ponytail Audit scans a repository for over-engineering and returns a ranked list of what to delete, simplify, or replace with standard library or native features. Use it before a code review to identify dead code, unnecessary abstractions, and duplicated platform functionality.
Quality
101.7K5.6K
✂️
3w ago

Ponytail Audit

Ponytail Audit scans a whole repository for over-engineering and ranks findings—dead code, standard-library or native replacements, YAGNI abstractions, and shrink opportunities—so you can decide what to delete or simplify. Use it when asked to audit a codebase, find bloat, or identify removable complexity.
Quality
101.7K5.6K
🔍
3w ago

Ponytail Debt

Ponytail Debt scans a codebase for `ponytail:` comments and reports them as a debt ledger, so deferred shortcuts are tracked instead of forgotten, and flags markers with no upgrade trigger.
Quality
101.7K5.6K
📒
3w ago

Ponytail Debt

Ponytail Debt scans a codebase for `ponytail:` comments and compiles them into a debt ledger, flagging entries with no trigger as rot risks. Use it when asked to list shortcuts or review what was deferred, as a one-shot report that changes nothing.
Quality
101.7K5.6K